A Survey regarding the Reporting of Simulation Studies

Hoaglin and Andrews (1975) proposed standards for computational practice and the reporting of computation-based studies.They observed that .statisticians . often pay too little attention to their own principles of design..To see if the design and reporting had improved, we surveyed five major statistical journals for 1975, 1978, and 1981 to ascertain whether reported simulation studies involved a specified design, justified the choice of the number of iterations, and specified the random number generator(s) used. Eighteen percent of the 1,198 papers surveyed included results based on simulation.We found that 9% of the papers including a simulation study justified the choice of the number of iterations and 44% at least partially specified the random number generator.Hoaglin and Andrews's observation appears still to be true.